Despite the increasing economic benefits generated by the cluster effect of the automotive industry, CO2 emissions have become a serious challenge in the “3060” dual-carbon context. There are significant uncertainties regarding the impact of automotive industry agglomeration on urban CO2 emissions, including quantifying impact thresholds and understanding the sensitivity of different agglomeration patterns. This study explores how automotive industry agglomeration impacts CO2 emissions across 278 Chinese cities, and uncover the impact patterns based on varying city characteristics. The results show that the technology agglomeration of the automobile industry exhibits a nonlinear “U-shaped” impact on CO2 emissions with a threshold value of 0.4267, while the scale agglomeration shows a nonlinear inverted “U-shaped” impact with a threshold value of 0.1206. In low-carbon demonstration cities, technology agglomeration has the most significant impact, with fluctuations ranging from −14.65% to 16.12%. In contrast, transitional cities are most affected by scale agglomeration, with a fluctuation range of −18.1%–11.19%. Besides, the results indicate that China's automobile industry development programs tend to relocate economic activities from highly agglomerated low-carbon demonstration cities to potential development cities with lower levels of agglomeration and other transitioning cities. Lastly, the results reconfirm that developing automotive industry development models based on cities' characteristics can enhance the environmental benefits of industrial agglomeration.